Jobst and I want to improve AI-safety by supplementing RLHF with a consensus generating voting system. Last week we did a small experiment at a conference. Here is the poster we used to explain this idea to the attendants:
Here’s the PDF
Jobst and I want to improve AI-safety by supplementing RLHF with a consensus generating voting system. Last week we did a small experiment at a conference. Here is the poster we used to explain this idea to the attendants:
Here’s the PDF